{
  "best_metric": 0.825,
  "best_model_checkpoint": "MAE-CT-CPC-Dicotomized-v8-n0-m1/checkpoint-1020",
  "epoch": 49.0004,
  "eval_steps": 500,
  "global_step": 2500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004,
      "grad_norm": 3.163785934448242,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.6803,
      "step": 10
    },
    {
      "epoch": 0.008,
      "grad_norm": 1.8414620161056519,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.6811,
      "step": 20
    },
    {
      "epoch": 0.012,
      "grad_norm": 2.732853412628174,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.6724,
      "step": 30
    },
    {
      "epoch": 0.016,
      "grad_norm": 2.302961826324463,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.6625,
      "step": 40
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.289227247238159,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.6826,
      "step": 50
    },
    {
      "epoch": 0.0204,
      "eval_accuracy": 0.7625,
      "eval_loss": 0.6156964302062988,
      "eval_runtime": 16.4829,
      "eval_samples_per_second": 4.854,
      "eval_steps_per_second": 1.213,
      "step": 51
    },
    {
      "epoch": 1.0036,
      "grad_norm": 2.539074420928955,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.653,
      "step": 60
    },
    {
      "epoch": 1.0076,
      "grad_norm": 4.338445663452148,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.6605,
      "step": 70
    },
    {
      "epoch": 1.0116,
      "grad_norm": 4.448472023010254,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.6647,
      "step": 80
    },
    {
      "epoch": 1.0156,
      "grad_norm": 8.590893745422363,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.665,
      "step": 90
    },
    {
      "epoch": 1.0196,
      "grad_norm": 4.512553691864014,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.6549,
      "step": 100
    },
    {
      "epoch": 1.0204,
      "eval_accuracy": 0.7625,
      "eval_loss": 0.5930413603782654,
      "eval_runtime": 15.1042,
      "eval_samples_per_second": 5.297,
      "eval_steps_per_second": 1.324,
      "step": 102
    },
    {
      "epoch": 2.0032,
      "grad_norm": 9.520898818969727,
      "learning_rate": 4.4e-06,
      "loss": 0.6033,
      "step": 110
    },
    {
      "epoch": 2.0072,
      "grad_norm": 7.96126651763916,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.5796,
      "step": 120
    },
    {
      "epoch": 2.0112,
      "grad_norm": 10.471980094909668,
      "learning_rate": 5.2e-06,
      "loss": 0.6731,
      "step": 130
    },
    {
      "epoch": 2.0152,
      "grad_norm": 17.355741500854492,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.6853,
      "step": 140
    },
    {
      "epoch": 2.0192,
      "grad_norm": 7.118047714233398,
      "learning_rate": 6e-06,
      "loss": 0.6486,
      "step": 150
    },
    {
      "epoch": 2.0204,
      "eval_accuracy": 0.5125,
      "eval_loss": 0.6796671152114868,
      "eval_runtime": 15.1483,
      "eval_samples_per_second": 5.281,
      "eval_steps_per_second": 1.32,
      "step": 153
    },
    {
      "epoch": 3.0028,
      "grad_norm": 9.448248863220215,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.627,
      "step": 160
    },
    {
      "epoch": 3.0068,
      "grad_norm": 10.044306755065918,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.5988,
      "step": 170
    },
    {
      "epoch": 3.0108,
      "grad_norm": 11.673822402954102,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.5388,
      "step": 180
    },
    {
      "epoch": 3.0148,
      "grad_norm": 13.179671287536621,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.4922,
      "step": 190
    },
    {
      "epoch": 3.0188,
      "grad_norm": 17.30994987487793,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.5595,
      "step": 200
    },
    {
      "epoch": 3.0204,
      "eval_accuracy": 0.6625,
      "eval_loss": 0.49017828702926636,
      "eval_runtime": 14.3197,
      "eval_samples_per_second": 5.587,
      "eval_steps_per_second": 1.397,
      "step": 204
    },
    {
      "epoch": 4.0024,
      "grad_norm": 12.384488105773926,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.5512,
      "step": 210
    },
    {
      "epoch": 4.0064,
      "grad_norm": 16.830211639404297,
      "learning_rate": 8.8e-06,
      "loss": 0.5532,
      "step": 220
    },
    {
      "epoch": 4.0104,
      "grad_norm": 25.703161239624023,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.4503,
      "step": 230
    },
    {
      "epoch": 4.0144,
      "grad_norm": 31.191774368286133,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.4281,
      "step": 240
    },
    {
      "epoch": 4.0184,
      "grad_norm": 7.301267147064209,
      "learning_rate": 1e-05,
      "loss": 0.5586,
      "step": 250
    },
    {
      "epoch": 4.0204,
      "eval_accuracy": 0.475,
      "eval_loss": 0.8194777369499207,
      "eval_runtime": 15.5636,
      "eval_samples_per_second": 5.14,
      "eval_steps_per_second": 1.285,
      "step": 255
    },
    {
      "epoch": 5.002,
      "grad_norm": 6.1059746742248535,
      "learning_rate": 9.955555555555556e-06,
      "loss": 0.4667,
      "step": 260
    },
    {
      "epoch": 5.006,
      "grad_norm": 11.250345230102539,
      "learning_rate": 9.911111111111113e-06,
      "loss": 0.3866,
      "step": 270
    },
    {
      "epoch": 5.01,
      "grad_norm": 11.811776161193848,
      "learning_rate": 9.866666666666668e-06,
      "loss": 0.3093,
      "step": 280
    },
    {
      "epoch": 5.014,
      "grad_norm": 2.898249387741089,
      "learning_rate": 9.822222222222223e-06,
      "loss": 0.4655,
      "step": 290
    },
    {
      "epoch": 5.018,
      "grad_norm": 39.88842010498047,
      "learning_rate": 9.777777777777779e-06,
      "loss": 0.4565,
      "step": 300
    },
    {
      "epoch": 5.0204,
      "eval_accuracy": 0.75,
      "eval_loss": 0.5872394442558289,
      "eval_runtime": 15.5445,
      "eval_samples_per_second": 5.147,
      "eval_steps_per_second": 1.287,
      "step": 306
    },
    {
      "epoch": 6.0016,
      "grad_norm": 18.96799659729004,
      "learning_rate": 9.733333333333334e-06,
      "loss": 0.6334,
      "step": 310
    },
    {
      "epoch": 6.0056,
      "grad_norm": 26.40775489807129,
      "learning_rate": 9.688888888888889e-06,
      "loss": 0.5505,
      "step": 320
    },
    {
      "epoch": 6.0096,
      "grad_norm": 23.987438201904297,
      "learning_rate": 9.644444444444444e-06,
      "loss": 0.4702,
      "step": 330
    },
    {
      "epoch": 6.0136,
      "grad_norm": 11.874549865722656,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.5567,
      "step": 340
    },
    {
      "epoch": 6.0176,
      "grad_norm": 1.870789647102356,
      "learning_rate": 9.555555555555556e-06,
      "loss": 0.3697,
      "step": 350
    },
    {
      "epoch": 6.0204,
      "eval_accuracy": 0.775,
      "eval_loss": 0.5016795992851257,
      "eval_runtime": 15.1939,
      "eval_samples_per_second": 5.265,
      "eval_steps_per_second": 1.316,
      "step": 357
    },
    {
      "epoch": 7.0012,
      "grad_norm": 1.9288091659545898,
      "learning_rate": 9.511111111111112e-06,
      "loss": 0.3384,
      "step": 360
    },
    {
      "epoch": 7.0052,
      "grad_norm": 47.43910598754883,
      "learning_rate": 9.466666666666667e-06,
      "loss": 0.4776,
      "step": 370
    },
    {
      "epoch": 7.0092,
      "grad_norm": 24.75432586669922,
      "learning_rate": 9.422222222222222e-06,
      "loss": 0.3272,
      "step": 380
    },
    {
      "epoch": 7.0132,
      "grad_norm": 6.029924392700195,
      "learning_rate": 9.377777777777779e-06,
      "loss": 0.6358,
      "step": 390
    },
    {
      "epoch": 7.0172,
      "grad_norm": 36.66233825683594,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.6201,
      "step": 400
    },
    {
      "epoch": 7.0204,
      "eval_accuracy": 0.7,
      "eval_loss": 0.6555034518241882,
      "eval_runtime": 15.1733,
      "eval_samples_per_second": 5.272,
      "eval_steps_per_second": 1.318,
      "step": 408
    },
    {
      "epoch": 8.0008,
      "grad_norm": 13.244248390197754,
      "learning_rate": 9.28888888888889e-06,
      "loss": 0.2307,
      "step": 410
    },
    {
      "epoch": 8.0048,
      "grad_norm": 0.1510910540819168,
      "learning_rate": 9.244444444444445e-06,
      "loss": 0.2198,
      "step": 420
    },
    {
      "epoch": 8.0088,
      "grad_norm": 48.222145080566406,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.1961,
      "step": 430
    },
    {
      "epoch": 8.0128,
      "grad_norm": 12.182064056396484,
      "learning_rate": 9.155555555555557e-06,
      "loss": 0.4113,
      "step": 440
    },
    {
      "epoch": 8.0168,
      "grad_norm": 90.49524688720703,
      "learning_rate": 9.111111111111112e-06,
      "loss": 0.4333,
      "step": 450
    },
    {
      "epoch": 8.0204,
      "eval_accuracy": 0.6125,
      "eval_loss": 1.2277292013168335,
      "eval_runtime": 15.8873,
      "eval_samples_per_second": 5.035,
      "eval_steps_per_second": 1.259,
      "step": 459
    },
    {
      "epoch": 9.0004,
      "grad_norm": 8.979384422302246,
      "learning_rate": 9.066666666666667e-06,
      "loss": 0.2832,
      "step": 460
    },
    {
      "epoch": 9.0044,
      "grad_norm": 124.57083892822266,
      "learning_rate": 9.022222222222223e-06,
      "loss": 0.4698,
      "step": 470
    },
    {
      "epoch": 9.0084,
      "grad_norm": 1.4499937295913696,
      "learning_rate": 8.977777777777778e-06,
      "loss": 0.622,
      "step": 480
    },
    {
      "epoch": 9.0124,
      "grad_norm": 204.4443359375,
      "learning_rate": 8.933333333333333e-06,
      "loss": 0.4289,
      "step": 490
    },
    {
      "epoch": 9.0164,
      "grad_norm": 13.056779861450195,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.3958,
      "step": 500
    },
    {
      "epoch": 9.0204,
      "grad_norm": 3.7403643131256104,
      "learning_rate": 8.844444444444445e-06,
      "loss": 0.2148,
      "step": 510
    },
    {
      "epoch": 9.0204,
      "eval_accuracy": 0.7625,
      "eval_loss": 0.8114517331123352,
      "eval_runtime": 14.7863,
      "eval_samples_per_second": 5.41,
      "eval_steps_per_second": 1.353,
      "step": 510
    },
    {
      "epoch": 10.004,
      "grad_norm": 0.3079785704612732,
      "learning_rate": 8.8e-06,
      "loss": 0.4126,
      "step": 520
    },
    {
      "epoch": 10.008,
      "grad_norm": 10.519120216369629,
      "learning_rate": 8.755555555555556e-06,
      "loss": 0.2208,
      "step": 530
    },
    {
      "epoch": 10.012,
      "grad_norm": 37.70737075805664,
      "learning_rate": 8.711111111111111e-06,
      "loss": 0.4088,
      "step": 540
    },
    {
      "epoch": 10.016,
      "grad_norm": 27.64992332458496,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.6548,
      "step": 550
    },
    {
      "epoch": 10.02,
      "grad_norm": 1.9343359470367432,
      "learning_rate": 8.622222222222223e-06,
      "loss": 0.9458,
      "step": 560
    },
    {
      "epoch": 10.0204,
      "eval_accuracy": 0.6625,
      "eval_loss": 0.9872623682022095,
      "eval_runtime": 14.8048,
      "eval_samples_per_second": 5.404,
      "eval_steps_per_second": 1.351,
      "step": 561
    },
    {
      "epoch": 11.0036,
      "grad_norm": 127.26286315917969,
      "learning_rate": 8.577777777777778e-06,
      "loss": 0.4077,
      "step": 570
    },
    {
      "epoch": 11.0076,
      "grad_norm": 48.36285400390625,
      "learning_rate": 8.533333333333335e-06,
      "loss": 0.315,
      "step": 580
    },
    {
      "epoch": 11.0116,
      "grad_norm": 1.3444968461990356,
      "learning_rate": 8.48888888888889e-06,
      "loss": 0.3529,
      "step": 590
    },
    {
      "epoch": 11.0156,
      "grad_norm": 1.425925374031067,
      "learning_rate": 8.444444444444446e-06,
      "loss": 0.2537,
      "step": 600
    },
    {
      "epoch": 11.0196,
      "grad_norm": 0.11205915361642838,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.0651,
      "step": 610
    },
    {
      "epoch": 11.0204,
      "eval_accuracy": 0.7625,
      "eval_loss": 1.0840221643447876,
      "eval_runtime": 16.1692,
      "eval_samples_per_second": 4.948,
      "eval_steps_per_second": 1.237,
      "step": 612
    },
    {
      "epoch": 12.0032,
      "grad_norm": 1.6690884828567505,
      "learning_rate": 8.355555555555556e-06,
      "loss": 0.0219,
      "step": 620
    },
    {
      "epoch": 12.0072,
      "grad_norm": 0.04994206875562668,
      "learning_rate": 8.311111111111111e-06,
      "loss": 0.3325,
      "step": 630
    },
    {
      "epoch": 12.0112,
      "grad_norm": 1.1334573030471802,
      "learning_rate": 8.266666666666667e-06,
      "loss": 0.1766,
      "step": 640
    },
    {
      "epoch": 12.0152,
      "grad_norm": 84.09977722167969,
      "learning_rate": 8.222222222222222e-06,
      "loss": 0.3474,
      "step": 650
    },
    {
      "epoch": 12.0192,
      "grad_norm": 0.11686452478170395,
      "learning_rate": 8.177777777777779e-06,
      "loss": 0.5756,
      "step": 660
    },
    {
      "epoch": 12.0204,
      "eval_accuracy": 0.8,
      "eval_loss": 1.0489223003387451,
      "eval_runtime": 14.837,
      "eval_samples_per_second": 5.392,
      "eval_steps_per_second": 1.348,
      "step": 663
    },
    {
      "epoch": 13.0028,
      "grad_norm": 0.4879148304462433,
      "learning_rate": 8.133333333333334e-06,
      "loss": 0.1627,
      "step": 670
    },
    {
      "epoch": 13.0068,
      "grad_norm": 10.051454544067383,
      "learning_rate": 8.08888888888889e-06,
      "loss": 0.1414,
      "step": 680
    },
    {
      "epoch": 13.0108,
      "grad_norm": 317.5495910644531,
      "learning_rate": 8.044444444444444e-06,
      "loss": 0.4045,
      "step": 690
    },
    {
      "epoch": 13.0148,
      "grad_norm": 12.827858924865723,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.2213,
      "step": 700
    },
    {
      "epoch": 13.0188,
      "grad_norm": 7.5428338050842285,
      "learning_rate": 7.955555555555557e-06,
      "loss": 0.354,
      "step": 710
    },
    {
      "epoch": 13.0204,
      "eval_accuracy": 0.7875,
      "eval_loss": 1.1601030826568604,
      "eval_runtime": 14.793,
      "eval_samples_per_second": 5.408,
      "eval_steps_per_second": 1.352,
      "step": 714
    },
    {
      "epoch": 14.0024,
      "grad_norm": 1.0718170404434204,
      "learning_rate": 7.911111111111112e-06,
      "loss": 0.1341,
      "step": 720
    },
    {
      "epoch": 14.0064,
      "grad_norm": 204.27011108398438,
      "learning_rate": 7.866666666666667e-06,
      "loss": 0.2009,
      "step": 730
    },
    {
      "epoch": 14.0104,
      "grad_norm": 623.6522827148438,
      "learning_rate": 7.822222222222224e-06,
      "loss": 0.2302,
      "step": 740
    },
    {
      "epoch": 14.0144,
      "grad_norm": 53.07473373413086,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.4521,
      "step": 750
    },
    {
      "epoch": 14.0184,
      "grad_norm": 0.032519057393074036,
      "learning_rate": 7.733333333333334e-06,
      "loss": 0.2888,
      "step": 760
    },
    {
      "epoch": 14.0204,
      "eval_accuracy": 0.625,
      "eval_loss": 1.8143768310546875,
      "eval_runtime": 15.8798,
      "eval_samples_per_second": 5.038,
      "eval_steps_per_second": 1.259,
      "step": 765
    },
    {
      "epoch": 15.002,
      "grad_norm": 457.6705322265625,
      "learning_rate": 7.68888888888889e-06,
      "loss": 0.2657,
      "step": 770
    },
    {
      "epoch": 15.006,
      "grad_norm": 431.8014221191406,
      "learning_rate": 7.644444444444445e-06,
      "loss": 0.0887,
      "step": 780
    },
    {
      "epoch": 15.01,
      "grad_norm": 0.013111516833305359,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.0183,
      "step": 790
    },
    {
      "epoch": 15.014,
      "grad_norm": 0.08553914725780487,
      "learning_rate": 7.555555555555556e-06,
      "loss": 0.0991,
      "step": 800
    },
    {
      "epoch": 15.018,
      "grad_norm": 0.11689701676368713,
      "learning_rate": 7.511111111111111e-06,
      "loss": 0.2449,
      "step": 810
    },
    {
      "epoch": 15.0204,
      "eval_accuracy": 0.7125,
      "eval_loss": 1.3988301753997803,
      "eval_runtime": 14.9719,
      "eval_samples_per_second": 5.343,
      "eval_steps_per_second": 1.336,
      "step": 816
    },
    {
      "epoch": 16.0016,
      "grad_norm": 0.12991830706596375,
      "learning_rate": 7.4666666666666675e-06,
      "loss": 0.009,
      "step": 820
    },
    {
      "epoch": 16.0056,
      "grad_norm": 185.4541015625,
      "learning_rate": 7.422222222222223e-06,
      "loss": 0.2772,
      "step": 830
    },
    {
      "epoch": 16.0096,
      "grad_norm": 0.7622888088226318,
      "learning_rate": 7.377777777777778e-06,
      "loss": 0.1701,
      "step": 840
    },
    {
      "epoch": 16.0136,
      "grad_norm": 0.5893406867980957,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.1631,
      "step": 850
    },
    {
      "epoch": 16.0176,
      "grad_norm": 0.7813571691513062,
      "learning_rate": 7.28888888888889e-06,
      "loss": 0.1326,
      "step": 860
    },
    {
      "epoch": 16.0204,
      "eval_accuracy": 0.7125,
      "eval_loss": 1.715152382850647,
      "eval_runtime": 15.1438,
      "eval_samples_per_second": 5.283,
      "eval_steps_per_second": 1.321,
      "step": 867
    },
    {
      "epoch": 17.0012,
      "grad_norm": 0.07987383008003235,
      "learning_rate": 7.244444444444445e-06,
      "loss": 0.0021,
      "step": 870
    },
    {
      "epoch": 17.0052,
      "grad_norm": 17.579423904418945,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.2835,
      "step": 880
    },
    {
      "epoch": 17.0092,
      "grad_norm": 0.05407591536641121,
      "learning_rate": 7.155555555555556e-06,
      "loss": 0.0557,
      "step": 890
    },
    {
      "epoch": 17.0132,
      "grad_norm": 129.4159393310547,
      "learning_rate": 7.111111111111112e-06,
      "loss": 0.147,
      "step": 900
    },
    {
      "epoch": 17.0172,
      "grad_norm": 0.6726216673851013,
      "learning_rate": 7.066666666666667e-06,
      "loss": 0.0018,
      "step": 910
    },
    {
      "epoch": 17.0204,
      "eval_accuracy": 0.6375,
      "eval_loss": 2.1475367546081543,
      "eval_runtime": 15.0746,
      "eval_samples_per_second": 5.307,
      "eval_steps_per_second": 1.327,
      "step": 918
    },
    {
      "epoch": 18.0008,
      "grad_norm": 0.06342015415430069,
      "learning_rate": 7.022222222222222e-06,
      "loss": 0.2025,
      "step": 920
    },
    {
      "epoch": 18.0048,
      "grad_norm": 0.03912360593676567,
      "learning_rate": 6.977777777777779e-06,
      "loss": 0.0017,
      "step": 930
    },
    {
      "epoch": 18.0088,
      "grad_norm": 0.015669086948037148,
      "learning_rate": 6.9333333333333344e-06,
      "loss": 0.1518,
      "step": 940
    },
    {
      "epoch": 18.0128,
      "grad_norm": 0.03878331929445267,
      "learning_rate": 6.88888888888889e-06,
      "loss": 0.6683,
      "step": 950
    },
    {
      "epoch": 18.0168,
      "grad_norm": 0.009163687005639076,
      "learning_rate": 6.844444444444445e-06,
      "loss": 0.3631,
      "step": 960
    },
    {
      "epoch": 18.0204,
      "eval_accuracy": 0.65,
      "eval_loss": 1.8957328796386719,
      "eval_runtime": 14.5069,
      "eval_samples_per_second": 5.515,
      "eval_steps_per_second": 1.379,
      "step": 969
    },
    {
      "epoch": 19.0004,
      "grad_norm": 0.030928779393434525,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.1663,
      "step": 970
    },
    {
      "epoch": 19.0044,
      "grad_norm": 0.8263186812400818,
      "learning_rate": 6.755555555555556e-06,
      "loss": 0.2357,
      "step": 980
    },
    {
      "epoch": 19.0084,
      "grad_norm": 0.09255637228488922,
      "learning_rate": 6.711111111111111e-06,
      "loss": 0.0141,
      "step": 990
    },
    {
      "epoch": 19.0124,
      "grad_norm": 0.10079475492238998,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.001,
      "step": 1000
    },
    {
      "epoch": 19.0164,
      "grad_norm": 0.1445166915655136,
      "learning_rate": 6.6222222222222236e-06,
      "loss": 0.1313,
      "step": 1010
    },
    {
      "epoch": 19.0204,
      "grad_norm": 0.0298333577811718,
      "learning_rate": 6.577777777777779e-06,
      "loss": 0.1252,
      "step": 1020
    },
    {
      "epoch": 19.0204,
      "eval_accuracy": 0.825,
      "eval_loss": 1.124619960784912,
      "eval_runtime": 14.4479,
      "eval_samples_per_second": 5.537,
      "eval_steps_per_second": 1.384,
      "step": 1020
    },
    {
      "epoch": 20.004,
      "grad_norm": 0.19592173397541046,
      "learning_rate": 6.533333333333334e-06,
      "loss": 0.0141,
      "step": 1030
    },
    {
      "epoch": 20.008,
      "grad_norm": 42.12409591674805,
      "learning_rate": 6.488888888888889e-06,
      "loss": 0.1769,
      "step": 1040
    },
    {
      "epoch": 20.012,
      "grad_norm": 482.6688537597656,
      "learning_rate": 6.444444444444445e-06,
      "loss": 0.1202,
      "step": 1050
    },
    {
      "epoch": 20.016,
      "grad_norm": 0.5874069333076477,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.0007,
      "step": 1060
    },
    {
      "epoch": 20.02,
      "grad_norm": 0.018259378150105476,
      "learning_rate": 6.355555555555556e-06,
      "loss": 0.0943,
      "step": 1070
    },
    {
      "epoch": 20.0204,
      "eval_accuracy": 0.6625,
      "eval_loss": 1.9498172998428345,
      "eval_runtime": 14.4129,
      "eval_samples_per_second": 5.551,
      "eval_steps_per_second": 1.388,
      "step": 1071
    },
    {
      "epoch": 21.0036,
      "grad_norm": 0.00311831571161747,
      "learning_rate": 6.311111111111111e-06,
      "loss": 0.2018,
      "step": 1080
    },
    {
      "epoch": 21.0076,
      "grad_norm": 7.55759859085083,
      "learning_rate": 6.266666666666668e-06,
      "loss": 0.333,
      "step": 1090
    },
    {
      "epoch": 21.0116,
      "grad_norm": 0.13464294373989105,
      "learning_rate": 6.222222222222223e-06,
      "loss": 0.2933,
      "step": 1100
    },
    {
      "epoch": 21.0156,
      "grad_norm": 0.013299187645316124,
      "learning_rate": 6.177777777777778e-06,
      "loss": 0.0908,
      "step": 1110
    },
    {
      "epoch": 21.0196,
      "grad_norm": 28.606412887573242,
      "learning_rate": 6.133333333333334e-06,
      "loss": 0.3488,
      "step": 1120
    },
    {
      "epoch": 21.0204,
      "eval_accuracy": 0.7875,
      "eval_loss": 1.3456709384918213,
      "eval_runtime": 15.9369,
      "eval_samples_per_second": 5.02,
      "eval_steps_per_second": 1.255,
      "step": 1122
    },
    {
      "epoch": 22.0032,
      "grad_norm": 2.039261817932129,
      "learning_rate": 6.08888888888889e-06,
      "loss": 0.2154,
      "step": 1130
    },
    {
      "epoch": 22.0072,
      "grad_norm": 6.409753799438477,
      "learning_rate": 6.044444444444445e-06,
      "loss": 0.0051,
      "step": 1140
    },
    {
      "epoch": 22.0112,
      "grad_norm": 0.010469136759638786,
      "learning_rate": 6e-06,
      "loss": 0.1893,
      "step": 1150
    },
    {
      "epoch": 22.0152,
      "grad_norm": 0.022801605984568596,
      "learning_rate": 5.955555555555555e-06,
      "loss": 0.0428,
      "step": 1160
    },
    {
      "epoch": 22.0192,
      "grad_norm": 0.014003835618495941,
      "learning_rate": 5.911111111111112e-06,
      "loss": 0.0008,
      "step": 1170
    },
    {
      "epoch": 22.0204,
      "eval_accuracy": 0.7125,
      "eval_loss": 1.7872467041015625,
      "eval_runtime": 14.9328,
      "eval_samples_per_second": 5.357,
      "eval_steps_per_second": 1.339,
      "step": 1173
    },
    {
      "epoch": 23.0028,
      "grad_norm": 336.1969299316406,
      "learning_rate": 5.8666666666666675e-06,
      "loss": 0.5488,
      "step": 1180
    },
    {
      "epoch": 23.0068,
      "grad_norm": 0.009750437922775745,
      "learning_rate": 5.822222222222223e-06,
      "loss": 0.3029,
      "step": 1190
    },
    {
      "epoch": 23.0108,
      "grad_norm": 0.15668730437755585,
      "learning_rate": 5.777777777777778e-06,
      "loss": 0.6396,
      "step": 1200
    },
    {
      "epoch": 23.0148,
      "grad_norm": 0.019450828433036804,
      "learning_rate": 5.733333333333334e-06,
      "loss": 0.0974,
      "step": 1210
    },
    {
      "epoch": 23.0188,
      "grad_norm": 6.186660289764404,
      "learning_rate": 5.688888888888889e-06,
      "loss": 0.009,
      "step": 1220
    },
    {
      "epoch": 23.0204,
      "eval_accuracy": 0.75,
      "eval_loss": 1.5437147617340088,
      "eval_runtime": 14.8598,
      "eval_samples_per_second": 5.384,
      "eval_steps_per_second": 1.346,
      "step": 1224
    },
    {
      "epoch": 24.0024,
      "grad_norm": 0.012686701491475105,
      "learning_rate": 5.6444444444444445e-06,
      "loss": 0.0007,
      "step": 1230
    },
    {
      "epoch": 24.0064,
      "grad_norm": 0.008507036603987217,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.0796,
      "step": 1240
    },
    {
      "epoch": 24.0104,
      "grad_norm": 0.05275079980492592,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.0045,
      "step": 1250
    },
    {
      "epoch": 24.0144,
      "grad_norm": 0.020087506622076035,
      "learning_rate": 5.511111111111112e-06,
      "loss": 0.0007,
      "step": 1260
    },
    {
      "epoch": 24.0184,
      "grad_norm": 0.0037845964543521404,
      "learning_rate": 5.466666666666667e-06,
      "loss": 0.0274,
      "step": 1270
    },
    {
      "epoch": 24.0204,
      "eval_accuracy": 0.6875,
      "eval_loss": 1.9865350723266602,
      "eval_runtime": 15.8645,
      "eval_samples_per_second": 5.043,
      "eval_steps_per_second": 1.261,
      "step": 1275
    },
    {
      "epoch": 25.002,
      "grad_norm": 0.008797760121524334,
      "learning_rate": 5.422222222222223e-06,
      "loss": 0.1231,
      "step": 1280
    },
    {
      "epoch": 25.006,
      "grad_norm": 0.012985019944608212,
      "learning_rate": 5.3777777777777784e-06,
      "loss": 0.0003,
      "step": 1290
    },
    {
      "epoch": 25.01,
      "grad_norm": 0.02057729661464691,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.058,
      "step": 1300
    },
    {
      "epoch": 25.014,
      "grad_norm": 1.6732549667358398,
      "learning_rate": 5.288888888888889e-06,
      "loss": 0.002,
      "step": 1310
    },
    {
      "epoch": 25.018,
      "grad_norm": 0.006808862090110779,
      "learning_rate": 5.244444444444445e-06,
      "loss": 0.0004,
      "step": 1320
    },
    {
      "epoch": 25.0204,
      "eval_accuracy": 0.7625,
      "eval_loss": 1.5100300312042236,
      "eval_runtime": 15.0066,
      "eval_samples_per_second": 5.331,
      "eval_steps_per_second": 1.333,
      "step": 1326
    },
    {
      "epoch": 26.0016,
      "grad_norm": 0.011416507884860039,
      "learning_rate": 5.2e-06,
      "loss": 0.0003,
      "step": 1330
    },
    {
      "epoch": 26.0056,
      "grad_norm": 0.006314845755696297,
      "learning_rate": 5.155555555555556e-06,
      "loss": 0.0236,
      "step": 1340
    },
    {
      "epoch": 26.0096,
      "grad_norm": 1.8917161226272583,
      "learning_rate": 5.1111111111111115e-06,
      "loss": 0.0011,
      "step": 1350
    },
    {
      "epoch": 26.0136,
      "grad_norm": 0.015401429496705532,
      "learning_rate": 5.0666666666666676e-06,
      "loss": 0.0003,
      "step": 1360
    },
    {
      "epoch": 26.0176,
      "grad_norm": 0.2598150372505188,
      "learning_rate": 5.022222222222223e-06,
      "loss": 0.1007,
      "step": 1370
    },
    {
      "epoch": 26.0204,
      "eval_accuracy": 0.6875,
      "eval_loss": 1.959010124206543,
      "eval_runtime": 15.1089,
      "eval_samples_per_second": 5.295,
      "eval_steps_per_second": 1.324,
      "step": 1377
    },
    {
      "epoch": 27.0012,
      "grad_norm": 0.025213167071342468,
      "learning_rate": 4.977777777777778e-06,
      "loss": 0.0352,
      "step": 1380
    },
    {
      "epoch": 27.0052,
      "grad_norm": 0.17898155748844147,
      "learning_rate": 4.933333333333334e-06,
      "loss": 0.0003,
      "step": 1390
    },
    {
      "epoch": 27.0092,
      "grad_norm": 0.011180482804775238,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.3099,
      "step": 1400
    },
    {
      "epoch": 27.0132,
      "grad_norm": 0.003275972092524171,
      "learning_rate": 4.8444444444444446e-06,
      "loss": 0.0006,
      "step": 1410
    },
    {
      "epoch": 27.0172,
      "grad_norm": 0.10510570555925369,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.0006,
      "step": 1420
    },
    {
      "epoch": 27.0204,
      "eval_accuracy": 0.7125,
      "eval_loss": 1.8345705270767212,
      "eval_runtime": 15.6323,
      "eval_samples_per_second": 5.118,
      "eval_steps_per_second": 1.279,
      "step": 1428
    },
    {
      "epoch": 28.0008,
      "grad_norm": 0.00974931288510561,
      "learning_rate": 4.755555555555556e-06,
      "loss": 0.0003,
      "step": 1430
    },
    {
      "epoch": 28.0048,
      "grad_norm": 0.01707894168794155,
      "learning_rate": 4.711111111111111e-06,
      "loss": 0.0002,
      "step": 1440
    },
    {
      "epoch": 28.0088,
      "grad_norm": 0.007560590747743845,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.0004,
      "step": 1450
    },
    {
      "epoch": 28.0128,
      "grad_norm": 66.74208068847656,
      "learning_rate": 4.622222222222222e-06,
      "loss": 0.1921,
      "step": 1460
    },
    {
      "epoch": 28.0168,
      "grad_norm": 0.0025084693916141987,
      "learning_rate": 4.5777777777777785e-06,
      "loss": 0.0006,
      "step": 1470
    },
    {
      "epoch": 28.0204,
      "eval_accuracy": 0.825,
      "eval_loss": 1.4668537378311157,
      "eval_runtime": 14.5998,
      "eval_samples_per_second": 5.48,
      "eval_steps_per_second": 1.37,
      "step": 1479
    },
    {
      "epoch": 29.0004,
      "grad_norm": 0.3155474364757538,
      "learning_rate": 4.533333333333334e-06,
      "loss": 0.0002,
      "step": 1480
    },
    {
      "epoch": 29.0044,
      "grad_norm": 0.024095896631479263,
      "learning_rate": 4.488888888888889e-06,
      "loss": 0.0011,
      "step": 1490
    },
    {
      "epoch": 29.0084,
      "grad_norm": 0.00578249292448163,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.0002,
      "step": 1500
    },
    {
      "epoch": 29.0124,
      "grad_norm": 0.009000943042337894,
      "learning_rate": 4.4e-06,
      "loss": 0.0008,
      "step": 1510
    },
    {
      "epoch": 29.0164,
      "grad_norm": 0.3360608220100403,
      "learning_rate": 4.3555555555555555e-06,
      "loss": 0.0004,
      "step": 1520
    },
    {
      "epoch": 29.0204,
      "grad_norm": 0.007300488650798798,
      "learning_rate": 4.3111111111111115e-06,
      "loss": 0.0001,
      "step": 1530
    },
    {
      "epoch": 29.0204,
      "eval_accuracy": 0.7875,
      "eval_loss": 1.5396068096160889,
      "eval_runtime": 14.8018,
      "eval_samples_per_second": 5.405,
      "eval_steps_per_second": 1.351,
      "step": 1530
    },
    {
      "epoch": 30.004,
      "grad_norm": 0.0051333606243133545,
      "learning_rate": 4.266666666666668e-06,
      "loss": 0.0002,
      "step": 1540
    },
    {
      "epoch": 30.008,
      "grad_norm": 0.006649728864431381,
      "learning_rate": 4.222222222222223e-06,
      "loss": 0.0002,
      "step": 1550
    },
    {
      "epoch": 30.012,
      "grad_norm": 0.004679904319345951,
      "learning_rate": 4.177777777777778e-06,
      "loss": 0.0002,
      "step": 1560
    },
    {
      "epoch": 30.016,
      "grad_norm": 220.38076782226562,
      "learning_rate": 4.133333333333333e-06,
      "loss": 0.0777,
      "step": 1570
    },
    {
      "epoch": 30.02,
      "grad_norm": 0.004013615660369396,
      "learning_rate": 4.088888888888889e-06,
      "loss": 0.0002,
      "step": 1580
    },
    {
      "epoch": 30.0204,
      "eval_accuracy": 0.7875,
      "eval_loss": 1.571619987487793,
      "eval_runtime": 14.813,
      "eval_samples_per_second": 5.401,
      "eval_steps_per_second": 1.35,
      "step": 1581
    },
    {
      "epoch": 31.0036,
      "grad_norm": 0.049216415733098984,
      "learning_rate": 4.044444444444445e-06,
      "loss": 0.0002,
      "step": 1590
    },
    {
      "epoch": 31.0076,
      "grad_norm": 0.01534576527774334,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0001,
      "step": 1600
    },
    {
      "epoch": 31.0116,
      "grad_norm": 0.002917769132182002,
      "learning_rate": 3.955555555555556e-06,
      "loss": 0.0002,
      "step": 1610
    },
    {
      "epoch": 31.0156,
      "grad_norm": 0.002222651382908225,
      "learning_rate": 3.911111111111112e-06,
      "loss": 0.0002,
      "step": 1620
    },
    {
      "epoch": 31.0196,
      "grad_norm": 0.008118602447211742,
      "learning_rate": 3.866666666666667e-06,
      "loss": 0.0001,
      "step": 1630
    },
    {
      "epoch": 31.0204,
      "eval_accuracy": 0.7625,
      "eval_loss": 1.6614097356796265,
      "eval_runtime": 14.4379,
      "eval_samples_per_second": 5.541,
      "eval_steps_per_second": 1.385,
      "step": 1632
    },
    {
      "epoch": 32.0032,
      "grad_norm": 0.004146672319620848,
      "learning_rate": 3.8222222222222224e-06,
      "loss": 0.0001,
      "step": 1640
    },
    {
      "epoch": 32.0072,
      "grad_norm": 0.0034797696862369776,
      "learning_rate": 3.777777777777778e-06,
      "loss": 0.0001,
      "step": 1650
    },
    {
      "epoch": 32.0112,
      "grad_norm": 0.0042143468745052814,
      "learning_rate": 3.7333333333333337e-06,
      "loss": 0.0001,
      "step": 1660
    },
    {
      "epoch": 32.0152,
      "grad_norm": 0.054984163492918015,
      "learning_rate": 3.688888888888889e-06,
      "loss": 0.0001,
      "step": 1670
    },
    {
      "epoch": 32.0192,
      "grad_norm": 0.004838942550122738,
      "learning_rate": 3.644444444444445e-06,
      "loss": 0.0002,
      "step": 1680
    },
    {
      "epoch": 32.0204,
      "eval_accuracy": 0.7625,
      "eval_loss": 1.6355606317520142,
      "eval_runtime": 15.4442,
      "eval_samples_per_second": 5.18,
      "eval_steps_per_second": 1.295,
      "step": 1683
    },
    {
      "epoch": 33.0028,
      "grad_norm": 0.3461035192012787,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.0001,
      "step": 1690
    },
    {
      "epoch": 33.0068,
      "grad_norm": 0.002831398043781519,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0001,
      "step": 1700
    },
    {
      "epoch": 33.0108,
      "grad_norm": 0.06873564422130585,
      "learning_rate": 3.511111111111111e-06,
      "loss": 0.0001,
      "step": 1710
    },
    {
      "epoch": 33.0148,
      "grad_norm": 0.006833823397755623,
      "learning_rate": 3.4666666666666672e-06,
      "loss": 0.0002,
      "step": 1720
    },
    {
      "epoch": 33.0188,
      "grad_norm": 0.004828931763768196,
      "learning_rate": 3.4222222222222224e-06,
      "loss": 0.0001,
      "step": 1730
    },
    {
      "epoch": 33.0204,
      "eval_accuracy": 0.8,
      "eval_loss": 1.5730502605438232,
      "eval_runtime": 14.4769,
      "eval_samples_per_second": 5.526,
      "eval_steps_per_second": 1.382,
      "step": 1734
    },
    {
      "epoch": 34.0024,
      "grad_norm": 0.02358504943549633,
      "learning_rate": 3.377777777777778e-06,
      "loss": 0.0003,
      "step": 1740
    },
    {
      "epoch": 34.0064,
      "grad_norm": 0.0021904546301811934,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.0237,
      "step": 1750
    },
    {
      "epoch": 34.0104,
      "grad_norm": 0.004002240486443043,
      "learning_rate": 3.2888888888888894e-06,
      "loss": 0.0002,
      "step": 1760
    },
    {
      "epoch": 34.0144,
      "grad_norm": 0.00806827750056982,
      "learning_rate": 3.2444444444444446e-06,
      "loss": 0.0001,
      "step": 1770
    },
    {
      "epoch": 34.0184,
      "grad_norm": 0.004369418602436781,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.0001,
      "step": 1780
    },
    {
      "epoch": 34.0204,
      "eval_accuracy": 0.725,
      "eval_loss": 2.0019965171813965,
      "eval_runtime": 14.7093,
      "eval_samples_per_second": 5.439,
      "eval_steps_per_second": 1.36,
      "step": 1785
    },
    {
      "epoch": 35.002,
      "grad_norm": 0.004029570147395134,
      "learning_rate": 3.1555555555555555e-06,
      "loss": 0.0001,
      "step": 1790
    },
    {
      "epoch": 35.006,
      "grad_norm": 0.0854596495628357,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.0001,
      "step": 1800
    },
    {
      "epoch": 35.01,
      "grad_norm": 0.00881748553365469,
      "learning_rate": 3.066666666666667e-06,
      "loss": 0.0001,
      "step": 1810
    },
    {
      "epoch": 35.014,
      "grad_norm": 0.007664634846150875,
      "learning_rate": 3.0222222222222225e-06,
      "loss": 0.0001,
      "step": 1820
    },
    {
      "epoch": 35.018,
      "grad_norm": 0.004043503198772669,
      "learning_rate": 2.9777777777777777e-06,
      "loss": 0.0001,
      "step": 1830
    },
    {
      "epoch": 35.0204,
      "eval_accuracy": 0.75,
      "eval_loss": 1.888606071472168,
      "eval_runtime": 16.1561,
      "eval_samples_per_second": 4.952,
      "eval_steps_per_second": 1.238,
      "step": 1836
    },
    {
      "epoch": 36.0016,
      "grad_norm": 0.027049187570810318,
      "learning_rate": 2.9333333333333338e-06,
      "loss": 0.0001,
      "step": 1840
    },
    {
      "epoch": 36.0056,
      "grad_norm": 0.011712036095559597,
      "learning_rate": 2.888888888888889e-06,
      "loss": 0.0001,
      "step": 1850
    },
    {
      "epoch": 36.0096,
      "grad_norm": 0.0028739357367157936,
      "learning_rate": 2.8444444444444446e-06,
      "loss": 0.0001,
      "step": 1860
    },
    {
      "epoch": 36.0136,
      "grad_norm": 0.0029218129348009825,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.0001,
      "step": 1870
    },
    {
      "epoch": 36.0176,
      "grad_norm": 0.002428996842354536,
      "learning_rate": 2.755555555555556e-06,
      "loss": 0.0001,
      "step": 1880
    },
    {
      "epoch": 36.0204,
      "eval_accuracy": 0.75,
      "eval_loss": 1.8363139629364014,
      "eval_runtime": 14.1077,
      "eval_samples_per_second": 5.671,
      "eval_steps_per_second": 1.418,
      "step": 1887
    },
    {
      "epoch": 37.0012,
      "grad_norm": 0.001762293977662921,
      "learning_rate": 2.7111111111111116e-06,
      "loss": 0.0001,
      "step": 1890
    },
    {
      "epoch": 37.0052,
      "grad_norm": 0.010973138734698296,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.0001,
      "step": 1900
    },
    {
      "epoch": 37.0092,
      "grad_norm": 0.007034891285002232,
      "learning_rate": 2.6222222222222225e-06,
      "loss": 0.0001,
      "step": 1910
    },
    {
      "epoch": 37.0132,
      "grad_norm": 0.04831545799970627,
      "learning_rate": 2.577777777777778e-06,
      "loss": 0.0001,
      "step": 1920
    },
    {
      "epoch": 37.0172,
      "grad_norm": 0.003924284130334854,
      "learning_rate": 2.5333333333333338e-06,
      "loss": 0.0001,
      "step": 1930
    },
    {
      "epoch": 37.0204,
      "eval_accuracy": 0.7625,
      "eval_loss": 1.6848043203353882,
      "eval_runtime": 14.1172,
      "eval_samples_per_second": 5.667,
      "eval_steps_per_second": 1.417,
      "step": 1938
    },
    {
      "epoch": 38.0008,
      "grad_norm": 0.003881203942000866,
      "learning_rate": 2.488888888888889e-06,
      "loss": 0.0001,
      "step": 1940
    },
    {
      "epoch": 38.0048,
      "grad_norm": 0.002467320766299963,
      "learning_rate": 2.4444444444444447e-06,
      "loss": 0.0001,
      "step": 1950
    },
    {
      "epoch": 38.0088,
      "grad_norm": 0.004267244599759579,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.0001,
      "step": 1960
    },
    {
      "epoch": 38.0128,
      "grad_norm": 0.003334041452035308,
      "learning_rate": 2.3555555555555555e-06,
      "loss": 0.1433,
      "step": 1970
    },
    {
      "epoch": 38.0168,
      "grad_norm": 0.0034292838536202908,
      "learning_rate": 2.311111111111111e-06,
      "loss": 0.0001,
      "step": 1980
    },
    {
      "epoch": 38.0204,
      "eval_accuracy": 0.75,
      "eval_loss": 1.7187621593475342,
      "eval_runtime": 14.1135,
      "eval_samples_per_second": 5.668,
      "eval_steps_per_second": 1.417,
      "step": 1989
    },
    {
      "epoch": 39.0004,
      "grad_norm": 0.002690413035452366,
      "learning_rate": 2.266666666666667e-06,
      "loss": 0.0142,
      "step": 1990
    },
    {
      "epoch": 39.0044,
      "grad_norm": 0.003135059028863907,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.0001,
      "step": 2000
    },
    {
      "epoch": 39.0084,
      "grad_norm": 0.003898640163242817,
      "learning_rate": 2.1777777777777777e-06,
      "loss": 0.0001,
      "step": 2010
    },
    {
      "epoch": 39.0124,
      "grad_norm": 0.0050819204188883305,
      "learning_rate": 2.133333333333334e-06,
      "loss": 0.0001,
      "step": 2020
    },
    {
      "epoch": 39.0164,
      "grad_norm": 0.0032317114528268576,
      "learning_rate": 2.088888888888889e-06,
      "loss": 0.0412,
      "step": 2030
    },
    {
      "epoch": 39.0204,
      "grad_norm": 0.00116757582873106,
      "learning_rate": 2.0444444444444447e-06,
      "loss": 0.0001,
      "step": 2040
    },
    {
      "epoch": 39.0204,
      "eval_accuracy": 0.8,
      "eval_loss": 1.5820459127426147,
      "eval_runtime": 15.2596,
      "eval_samples_per_second": 5.243,
      "eval_steps_per_second": 1.311,
      "step": 2040
    },
    {
      "epoch": 40.004,
      "grad_norm": 0.00318012572824955,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0001,
      "step": 2050
    },
    {
      "epoch": 40.008,
      "grad_norm": 0.0023195173125714064,
      "learning_rate": 1.955555555555556e-06,
      "loss": 0.0008,
      "step": 2060
    },
    {
      "epoch": 40.012,
      "grad_norm": 0.002880761167034507,
      "learning_rate": 1.9111111111111112e-06,
      "loss": 0.0001,
      "step": 2070
    },
    {
      "epoch": 40.016,
      "grad_norm": 88.72091674804688,
      "learning_rate": 1.8666666666666669e-06,
      "loss": 0.0023,
      "step": 2080
    },
    {
      "epoch": 40.02,
      "grad_norm": 0.0025204592384397984,
      "learning_rate": 1.8222222222222225e-06,
      "loss": 0.0001,
      "step": 2090
    },
    {
      "epoch": 40.0204,
      "eval_accuracy": 0.7875,
      "eval_loss": 1.6061248779296875,
      "eval_runtime": 14.5072,
      "eval_samples_per_second": 5.515,
      "eval_steps_per_second": 1.379,
      "step": 2091
    },
    {
      "epoch": 41.0036,
      "grad_norm": 0.0022217093501240015,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.167,
      "step": 2100
    },
    {
      "epoch": 41.0076,
      "grad_norm": 0.0016050190897658467,
      "learning_rate": 1.7333333333333336e-06,
      "loss": 0.0001,
      "step": 2110
    },
    {
      "epoch": 41.0116,
      "grad_norm": 0.0031738209072500467,
      "learning_rate": 1.688888888888889e-06,
      "loss": 0.0001,
      "step": 2120
    },
    {
      "epoch": 41.0156,
      "grad_norm": 0.0036317266058176756,
      "learning_rate": 1.6444444444444447e-06,
      "loss": 0.0001,
      "step": 2130
    },
    {
      "epoch": 41.0196,
      "grad_norm": 0.002648336812853813,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.0001,
      "step": 2140
    },
    {
      "epoch": 41.0204,
      "eval_accuracy": 0.7,
      "eval_loss": 2.2816524505615234,
      "eval_runtime": 14.2804,
      "eval_samples_per_second": 5.602,
      "eval_steps_per_second": 1.401,
      "step": 2142
    },
    {
      "epoch": 42.0032,
      "grad_norm": 0.0022142785601317883,
      "learning_rate": 1.5555555555555558e-06,
      "loss": 0.0001,
      "step": 2150
    },
    {
      "epoch": 42.0072,
      "grad_norm": 0.006730781402438879,
      "learning_rate": 1.5111111111111112e-06,
      "loss": 0.0001,
      "step": 2160
    },
    {
      "epoch": 42.0112,
      "grad_norm": 0.0027248021215200424,
      "learning_rate": 1.4666666666666669e-06,
      "loss": 0.0001,
      "step": 2170
    },
    {
      "epoch": 42.0152,
      "grad_norm": 0.0026180455461144447,
      "learning_rate": 1.4222222222222223e-06,
      "loss": 0.0001,
      "step": 2180
    },
    {
      "epoch": 42.0192,
      "grad_norm": 0.0017348204273730516,
      "learning_rate": 1.377777777777778e-06,
      "loss": 0.0001,
      "step": 2190
    },
    {
      "epoch": 42.0204,
      "eval_accuracy": 0.725,
      "eval_loss": 2.101508617401123,
      "eval_runtime": 15.3729,
      "eval_samples_per_second": 5.204,
      "eval_steps_per_second": 1.301,
      "step": 2193
    },
    {
      "epoch": 43.0028,
      "grad_norm": 0.0021387911401689053,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0001,
      "step": 2200
    },
    {
      "epoch": 43.0068,
      "grad_norm": 0.0030123190954327583,
      "learning_rate": 1.288888888888889e-06,
      "loss": 0.0001,
      "step": 2210
    },
    {
      "epoch": 43.0108,
      "grad_norm": 0.0043581160716712475,
      "learning_rate": 1.2444444444444445e-06,
      "loss": 0.0001,
      "step": 2220
    },
    {
      "epoch": 43.0148,
      "grad_norm": 0.0023331050761044025,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.0001,
      "step": 2230
    },
    {
      "epoch": 43.0188,
      "grad_norm": 0.0014669563388451934,
      "learning_rate": 1.1555555555555556e-06,
      "loss": 0.0001,
      "step": 2240
    },
    {
      "epoch": 43.0204,
      "eval_accuracy": 0.775,
      "eval_loss": 1.6356258392333984,
      "eval_runtime": 14.3954,
      "eval_samples_per_second": 5.557,
      "eval_steps_per_second": 1.389,
      "step": 2244
    },
    {
      "epoch": 44.0024,
      "grad_norm": 0.0022818814031779766,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.2123,
      "step": 2250
    },
    {
      "epoch": 44.0064,
      "grad_norm": 0.004321521148085594,
      "learning_rate": 1.066666666666667e-06,
      "loss": 0.0001,
      "step": 2260
    },
    {
      "epoch": 44.0104,
      "grad_norm": 0.002616771264001727,
      "learning_rate": 1.0222222222222223e-06,
      "loss": 0.0001,
      "step": 2270
    },
    {
      "epoch": 44.0144,
      "grad_norm": 0.008771556429564953,
      "learning_rate": 9.77777777777778e-07,
      "loss": 0.0001,
      "step": 2280
    },
    {
      "epoch": 44.0184,
      "grad_norm": 0.007163883652538061,
      "learning_rate": 9.333333333333334e-07,
      "loss": 0.0001,
      "step": 2290
    },
    {
      "epoch": 44.0204,
      "eval_accuracy": 0.8125,
      "eval_loss": 1.5849277973175049,
      "eval_runtime": 14.7946,
      "eval_samples_per_second": 5.407,
      "eval_steps_per_second": 1.352,
      "step": 2295
    },
    {
      "epoch": 45.002,
      "grad_norm": 0.001609973143786192,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.0001,
      "step": 2300
    },
    {
      "epoch": 45.006,
      "grad_norm": 0.0015221175272017717,
      "learning_rate": 8.444444444444445e-07,
      "loss": 0.0001,
      "step": 2310
    },
    {
      "epoch": 45.01,
      "grad_norm": 0.00561766279861331,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.0002,
      "step": 2320
    },
    {
      "epoch": 45.014,
      "grad_norm": 0.005561948753893375,
      "learning_rate": 7.555555555555556e-07,
      "loss": 0.0001,
      "step": 2330
    },
    {
      "epoch": 45.018,
      "grad_norm": 0.0020426807459443808,
      "learning_rate": 7.111111111111112e-07,
      "loss": 0.0001,
      "step": 2340
    },
    {
      "epoch": 45.0204,
      "eval_accuracy": 0.775,
      "eval_loss": 1.6463369131088257,
      "eval_runtime": 15.349,
      "eval_samples_per_second": 5.212,
      "eval_steps_per_second": 1.303,
      "step": 2346
    },
    {
      "epoch": 46.0016,
      "grad_norm": 0.011843581683933735,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.0001,
      "step": 2350
    },
    {
      "epoch": 46.0056,
      "grad_norm": 0.003159413579851389,
      "learning_rate": 6.222222222222223e-07,
      "loss": 0.0001,
      "step": 2360
    },
    {
      "epoch": 46.0096,
      "grad_norm": 0.021830186247825623,
      "learning_rate": 5.777777777777778e-07,
      "loss": 0.0001,
      "step": 2370
    },
    {
      "epoch": 46.0136,
      "grad_norm": 0.0015857354737818241,
      "learning_rate": 5.333333333333335e-07,
      "loss": 0.0001,
      "step": 2380
    },
    {
      "epoch": 46.0176,
      "grad_norm": 0.0021360372193157673,
      "learning_rate": 4.88888888888889e-07,
      "loss": 0.0001,
      "step": 2390
    },
    {
      "epoch": 46.0204,
      "eval_accuracy": 0.775,
      "eval_loss": 1.664137601852417,
      "eval_runtime": 14.3393,
      "eval_samples_per_second": 5.579,
      "eval_steps_per_second": 1.395,
      "step": 2397
    },
    {
      "epoch": 47.0012,
      "grad_norm": 7.805647850036621,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.0007,
      "step": 2400
    },
    {
      "epoch": 47.0052,
      "grad_norm": 0.002305046422407031,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.0001,
      "step": 2410
    },
    {
      "epoch": 47.0092,
      "grad_norm": 0.0024446428287774324,
      "learning_rate": 3.555555555555556e-07,
      "loss": 0.0001,
      "step": 2420
    },
    {
      "epoch": 47.0132,
      "grad_norm": 0.0018226341344416142,
      "learning_rate": 3.111111111111111e-07,
      "loss": 0.0001,
      "step": 2430
    },
    {
      "epoch": 47.0172,
      "grad_norm": 0.0018344988347962499,
      "learning_rate": 2.666666666666667e-07,
      "loss": 0.0001,
      "step": 2440
    },
    {
      "epoch": 47.0204,
      "eval_accuracy": 0.7875,
      "eval_loss": 1.612348198890686,
      "eval_runtime": 14.4481,
      "eval_samples_per_second": 5.537,
      "eval_steps_per_second": 1.384,
      "step": 2448
    },
    {
      "epoch": 48.0008,
      "grad_norm": 0.0016147164860740304,
      "learning_rate": 2.2222222222222224e-07,
      "loss": 0.0001,
      "step": 2450
    },
    {
      "epoch": 48.0048,
      "grad_norm": 0.0015180219197645783,
      "learning_rate": 1.777777777777778e-07,
      "loss": 0.0001,
      "step": 2460
    },
    {
      "epoch": 48.0088,
      "grad_norm": 0.0027082718443125486,
      "learning_rate": 1.3333333333333336e-07,
      "loss": 0.0001,
      "step": 2470
    },
    {
      "epoch": 48.0128,
      "grad_norm": 0.0015189133118838072,
      "learning_rate": 8.88888888888889e-08,
      "loss": 0.0001,
      "step": 2480
    },
    {
      "epoch": 48.0168,
      "grad_norm": 0.0014700506580993533,
      "learning_rate": 4.444444444444445e-08,
      "loss": 0.0001,
      "step": 2490
    },
    {
      "epoch": 48.0204,
      "eval_accuracy": 0.7875,
      "eval_loss": 1.6145384311676025,
      "eval_runtime": 14.6364,
      "eval_samples_per_second": 5.466,
      "eval_steps_per_second": 1.366,
      "step": 2499
    },
    {
      "epoch": 49.0004,
      "grad_norm": 0.005640446674078703,
      "learning_rate": 0.0,
      "loss": 0.0001,
      "step": 2500
    },
    {
      "epoch": 49.0004,
      "eval_accuracy": 0.7875,
      "eval_loss": 1.6145387887954712,
      "eval_runtime": 15.4845,
      "eval_samples_per_second": 5.166,
      "eval_steps_per_second": 1.292,
      "step": 2500
    },
    {
      "epoch": 49.0004,
      "step": 2500,
      "total_flos": 4.34799425740686e+19,
      "train_loss": 0.1747144501608098,
      "train_runtime": 4798.9949,
      "train_samples_per_second": 2.084,
      "train_steps_per_second": 0.521
    },
    {
      "epoch": 49.0004,
      "eval_accuracy": 0.5753424657534246,
      "eval_loss": 2.5945777893066406,
      "eval_runtime": 15.5022,
      "eval_samples_per_second": 4.709,
      "eval_steps_per_second": 1.226,
      "step": 2500
    },
    {
      "epoch": 49.0004,
      "eval_accuracy": 0.5753424657534246,
      "eval_loss": 2.5945777893066406,
      "eval_runtime": 14.1475,
      "eval_samples_per_second": 5.16,
      "eval_steps_per_second": 1.343,
      "step": 2500
    }
  ],
  "logging_steps": 10,
  "max_steps": 2500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.34799425740686e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}